Articles

Clinical characteristics and outcomes of invasively ventilated

patients with COVID-19 in Argentina (SATICOVID):
a prospective, multicentre cohort study
Elisa Estenssoro, Cecilia I Loudet, Fernando G Ríos, Vanina S Kanoore Edul, Gustavo Plotnikow, Macarena Andrian, Ignacio Romero,
Damián Piezny, Marco Bezzi, Verónica Mandich, Carla Groer, Sebastián Torres, Cristina Orlandi, Paolo N Rubatto Birri, María F Valenti,
Eleonora Cunto, María G Sáenz, Norberto Tiribelli, Vanina Aphalo, Rosa Reina, Arnaldo Dubin on behalf of the SATI-COVID-19 Study Group*

Summary
Background Although COVID-19 has greatly affected many low-income and middle-income countries, detailed Lancet Respir Med 2021
information about patients admitted to the intensive care unit (ICU) is still scarce. Our aim was to examine ventilation Published Online
characteristics and outcomes in invasively ventilated patients with COVID-19 in Argentina, an upper middle-income July 2, 2021
https://doi.org/10.1016/
country.
S2213-2600(21)00229-0
See Online/Comment
Methods In this prospective, multicentre cohort study (SATICOVID), we enrolled patients aged 18 years or older with https://doi.org/10.1016/
RT-PCR-confirmed COVID-19 who were on invasive mechanical ventilation and admitted to one of 63 ICUs in S2213-2600(21)00267-8
Argentina. Patient demographics and clinical, laboratory, and general management variables were collected on day 1 See Online/Profile
(ICU admission); physiological respiratory and ventilation variables were collected on days 1, 3, and 7. The primary https://doi.org/10.1016/
S2213-2600(21)00321-0
outcome was all-cause in-hospital mortality. All patients were followed until death in hospital or hospital discharge,
whichever occurred first. Secondary outcomes were ICU mortality, identification of independent predictors of For the Spanish translation of the
Summary see Online for
mortality, duration of invasive mechanical ventilation, and patterns of change in physiological respiratory and appendix 1
mechanical ventilation variables. The study is registered with ClinicalTrials.gov, NCT04611269, and is complete. *Group members are listed in
appendix 2
Findings Between March 20, 2020, and Oct 31, 2020, we enrolled 1909 invasively ventilated patients with COVID-19, Hospital Interzonal de Agudos
with a median age of 62 years [IQR 52–70]. 1294 (67·8%) were men, hypertension and obesity were the main General San Martín, Buenos
comorbidities, and 939 (49·2%) patients required vasopressors. Lung-protective ventilation was widely used and Aires, Argentina
(E Estenssoro MD, C I Loudet MD,
median duration of ventilation was 13 days (IQR 7–22). Median tidal volume was 6∙1 mL/kg predicted bodyweight
M G Sáenz MD); Sanatorio Las
(IQR 6∙0–7∙0) on day 1, and the value increased significantly up to day 7; positive end-expiratory pressure was 10 cm Lomas, Buenos Aires, Argentina
H2O (8–12) on day 1, with a slight but significant decrease to day 7. Ratio of partial pressure of arterial oxygen (PaO2) (F G Ríos MD); Hospital
to fractional inspired oxygen (FiO2) was 160 (IQR 111–218), respiratory system compliance 36 mL/cm H2O (29–44), Juan A Fernández, Buenos Aires,
Argentina
driving pressure 12 cm H2O (10–14), and FiO2 0·60 (0·45–0·80) on day 1. Acute respiratory distress syndrome (V S Kanoore Edul PhD,
developed in 1672 (87·6%) of patients; 1176 (61·6%) received prone positioning. In-hospital mortality was C Groer MD); Sanatorio
57·7% (1101/1909 patients) and ICU mortality was 57∙0% (1088/1909 patients); 462 (43·8%) patients died of refractory Anchorena, Buenos Aires,
hypoxaemia, frequently overlapping with septic shock (n=174). Cox regression identified age (hazard ratio 1∙02 Argentina (G Plotnikow RT,
V Aphalo MD); Hospital
[95% CI 1∙01–1∙03]), Charlson score (1∙16 [1∙11–1∙23]), endotracheal intubation outside of the ICU (ie, before ICU Provincial Dr Castro Rendón,
admission; 1∙37 [1∙10–1∙71]), vasopressor use on day 1 (1∙29 [1∙07–1∙55]), D-dimer concentration (1∙02 [1∙01–1∙03]), Neuquén, Argentina
PaO2/FiO2 on day 1 (0∙998 [0∙997–0∙999]), arterial pH on day 1 (1∙01 [1∙00–1∙01]), driving pressure on day 1 (1∙05 (M Andrian MD); Sanatorio
[1∙03–1∙08]), acute kidney injury (1∙66 [1∙36–2∙03]), and month of admission (1∙10 [1∙03–1∙18]) as independent Güemes, Buenos Aires,
Argentina (I Romero MD);
predictors of mortality. Hospital A Posadas, Buenos
Aires, Argentina (D Piezny MD);
Interpretation In patients with COVID-19 who required invasive mechanical ventilation, lung-protective ventilation Hospital Santojanni, Buenos
was widely used but mortality was high. Predictors of mortality in our study broadly agreed with those identified in Aires, Argentina (V Mandich MD,
M Bezzi RT); Sanatorio
studies of invasively ventilated patients in high-income countries. The sustained burden of COVID-19 on scarce Anchorena San Martín, Buenos
health-care personnel might have contributed to high mortality over the course of our study in Argentina. These data Aires, Argentina (S Torres MD);
might help to identify points for improvement in the management of patients in middle-income countries and Hospital Francisco Lopez Lima,
elsewhere. Río Negro, Argentina
(C Orlandi MD); Sanatorio
Otamendi, Buenos Aires,
Funding None. Argentina (P N Rubatto Birri MD,
A Dubin PhD); Sanatorio de Los
Copyright Copyright © 2021 Elsevier Ltd. All rights reserved. Arcos, Buenos Aires, Argentina
(M F Valenti MD); Hospital
Dr F J Muñiz, Buenos Aires,
Introduction COVID-19 a pandemic; as of May 1, 2021, 153 480 005 cases Argentina (E Cunto MD);
Since the first case of pneumonia related to SARS-CoV-2 of COVID-19 had been confirmed, with 3 206 117 deaths.1 Complejo Médico de la Policía
Federal Argentina Churruca
was reported in 2019, COVID-19 has spread relentlessly From the beginning of the pandemic, there was great
Visca, Buenos Aires, Argentina
across the world. On March 11, 2020, WHO declared concern in the clinical and research communities about (N Tiribelli RT); Sociedad

www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0 1

 Articles
Argentina de Terapia Intensiva,
Buenos Aires, Argentina
(R Reina MD)
Correspondence to:
Dr Elisa Estenssoro, Servicio de
Terapia Intensiva, Hospital
Interzonal de Agudos General
San Martín, Calle 42 No. 577,
1900 La Plata, Provincia de
Buenos Aires, Argentina
estenssoro.elisa@gmail.com
See Online for appendix 2
For the SATI website see
https://www.sati.org.ar/
2 www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0
Research in context
Evidence before this study
Although the impact of COVID-19 on low-income and middle-
income countries (LMICs) is widely recognised , little is known
about outcomes for patients receiving invasive mechanical
ventilation in these regions. We searched PubMed on
Feb 12, 2021, for studies of adults patients (≥18 years) using the
terms “coronavirus” OR “COVID-19” AND “mechanical
ventilation” AND “ICU” AND “mortality”. We included articles that
had at least an abstract written in English. Our search identified
258 articles, many of which were single-centre studies with few
patients. We identified nationwide, retrospective cohort studies
that reported epidemiological characteristics and outcome
measures from Germany, Mexico, Brazil, Iran, the UK, and the
USA. The first three of these studies reported the number of
patients admitted to the intensive care unit (ICU), the number of
patients who required mechanical ventilation, and mortality;
but there was no mention of mechanical ventilation variables.
Mortality was greater than 55% in these studies. We also
identified another four multicentre cohort studies that included
patients on mechanical ventilation, which provide detailed data
analysis of mechanical ventilation variables, and the association
of these variables with outcomes. These studies were from the
Netherlands (553 patients from 18 ICUs); Italy (1591 patients
from 72 ICUs); France, Belgium, and Switzerland (the REVA
cohort, 3376 patients from 138 ICUs); and Spain (742 patients
from 36 ICUs, all with acute respiratory distress syndrome).
Only the REVA cohort and the study from Spain were prospective
studies. None of them originated in a LMIC.
its potential impact on low-income and middle-income
countries (LMICs), given their profound, long-lasting
economic and educational inequities, social turbulence,
and fragile health systems.2–4 Epidemio­ logical infor­
mation about critically ill patients with COVID-19 in
LMICs has been scarce, although some countries
with nationwide databases have reported worthwhile
information.5–7 In Argentina, an upper middle-income
country (defined by the World Bank as economies with a
gross national income per capita of between US$4046
and US$12 535), information provided by the Ministry
of Health is fragmented owing to the absence of an
integrated health-care system, and available data about
private health subsectors is deficient. In this context, the
Argentine Society of Intensive Care (Sociedad Argentina
de Terapia Intensiva or SATI) launched a prospective
cohort study with the aim of describing epidemiological,
clinical, and physio­ logical character­ istics, ventilation
settings and received treatm­ ents, and outcomes in
patients with laboratory-confirmed COVID-19 who
required invasive mechanical ventilation.
Our primary aim was to determine in-hospital
mortality. Given that hospital mortality for patients with
critical conditions such as sepsis, and for mechanically
ventilated patients with respiratory failure due to
Added value of this study
To our knowledge, SATICOVID is the first prospective,
multicentre cohort study carried out in a LMIC during the
pandemic; it includes 1909 patients with RT-PCR-confirmed
COVID-19 from 63 ICUs in Argentina. The study provides a
detailed analysis of patient epidemiological characteristics,
laboratory findings, symptoms, respiratory physiological
parameters, and mechanical ventilation variables over time,
and causes of death. The entire cohort of patients was followed
until death or hospital discharge.
Implications of all the available evidence
Lung-protective ventilation was widely used in Argentina,
as elsewhere. Overall in-hospital mortality for invasively
ventilated patients with COVID-19 was high, as has been
reported in other regions (eg, Mexico and Brazil). Mortality was
related to age, comorbidities, acute cardiovascular and kidney
dysfunction, and compromised oxygenation; we also found an
association with driving pressure, a variable of respiratory
mechanics. Although the health system in Argentina was well
resourced in terms of equipment and consumables during
periods of high demand, mortality increased throughout the
study period, from April to October, 2020, perhaps as a result of
the sustained burden on scarce health-care personnel.
Our findings add to the existing body of knowledge about
COVID-19 epidemiological aspects and outcomes, and also
about the current practice of mechanical ventilation.
2009 H1N1 influenza, was reported to be higher in
Argentina than in high-income countries,3,8 we hypo­
thesised that in invasively ventilated patients with
confirmed COVID-19, hospital mortality in our cohort
would be higher than the 26% reported for the Lombardy
Region, Italy,9 at the time of study initiation.
Methods
Study design and population
SATICOVID was a prospective, multicentre cohort study
that enrolled patients aged 18 years and older with
RT-PCR-confirmed SARS-CoV-2 infection who required
invasive mechanical ventilation and were admitted to
63 ICUs in Argentina (appendix 2, pp 5–6). As specified
in the protocol, patients were excluded from the analysis
if SARS-CoV-2 infection was not confirmed, according to
WHO guidance, or if they had a severe respiratory
infection or pneumonia proven to be due to another
cause. Patients were also excluded if no baseline data
were recorded or if no details of ventilatory parameters
were available. Patients were followed until death in
hospital or hospital discharge, whichever occurred first,
allowing a complete case analysis.
SATI announced the study on its website and via emails
to all society members to invite them to participate in the

study. Electronic forms for reporting of hospital and ICU
data, case characteristics, and ventilation parameters
were designed and distributed by the investigators via
email (appendix 2, pp 48–64). Paper forms for case
reporting were also available to local researchers to
complete, scan, and send to a specific email address.
These forms were provided by the principal investigator
(EE). The email was monitored by EE and VSKE. A
trained data-entry specialist (who was not involved in any
other part of the study) added the data collected from
these forms to a central database (Excel), which was then
was exported into a Stata dta. Only EE, AD, and CIL had
access to the database. Individual patient data were
anonymised by assigning a numerical code to each case.
Code numbers were assigned in order of admission.
Each local institutional review board approved the
study and defined the requirement for informed
consent. The SATICOVID study protocol is available in
appendix 2 (p 26).
Recorded variables
On day 1, at ICU admission, we recorded patient
demographics and characteristics, including date of
admission; age (as a continuous variable and in 10-year
categories: <40 years, 40–49 years, 50–59 years, and so on,
up to ≥80 years); sex; body-mass index (BMI); comor­
bidities and Charlson score; Acute Physiology and Chronic
Health Evaluation (APACHE) II and Sequential Organ
Failure Assessment (SOFA) scores; need for vasopressors;
laboratory variables; smoking status; and any alcohol-
related problem (appendix 2, p 34). We also registered
patient signs and symptoms of infection before hospital
admission; number of days from symptom onset to
hospital admission; number of days from hospital
admission to initiation of invasive mechanical ventilation;
preintubation use of high-flow nasal cannula (HFNC) and
non-invasive mechanical ventilation (NIV); and the site of
endotracheal intubation (outside or inside the ICU).
Physiological respiratory and mechanical ventilation
variables were collected on admission to the ICU (day 1)
and on days 3 and 7: blood gas analysis (arterial pH,
partial pressure of arterial carbon dioxide (PaCO2), partial
pressure of arterial oxygen (PaO2), and arterial oxygen
saturation; plasma bicarbonate and base excess were
then calculated); proportion of patients with lung
infiltrates involving 3–4 quadrants on chest x-ray; ratio of
PaO2 to fractional inspired oxygen (FiO2); tidal volume
in mL/kg predicted bodyweight; FiO2; respiratory rate;
positive end-expiratory pressure (PEEP); plateau
pressure; respiratory system compliance; and driving
pressure. Score on the Richmond Agitation-Sedation
Scale (RASS) was recorded, with values of –5 and
–4 points considered as deep sedation.
We also registered development of acute respiratory
distress syndrome (ARDS);10 use of prone positioning,
including number and duration of sessions (in hours);
acute kidney failure and requirement for renal
www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0 3
Articles
98 ICUs contacted across Argentina
17 declined or did not respond
81 agreed to participate
18 excluded
4 ICUs did not include patients
2 no IRB approval obtained
12 data collection not possible*
63 participating ICUs
4298 patients with PCR-confirmed COVID-19
2315 invasive ventilation not required
1983 patients with COVID-19 on invasive ventilation assessed for
eligibility
61 excluded
45 alternative diagnosis
16 no informed consent†
1922 patients with COVID-19 on invasive ventilation
12 insufficient data
1909 patients with COVID-19 on invasive ventilation with data
Figure 1: Study profile
IRB=institutional review board. *Limited capacity to enroll patients because of
insufficient staff resources during a time of high COVID-19 activity. †Each local
IRB defined the requirement for informed consent.
replacement therapy; septic shock; maximum fever
value; development of bacteraemia; ventilation-asso­
ciated pneumonia; thromboembolic events and their
local­
isation; and use of specific treatments. Cortico­
steroid use was analysed before and after the publication
of the RECOVERY trial, which showed a benefit of
dexamethasone in hospitalised patients with COVID-19.11
Causes of death were selected from a list of
nine predetermined possibilities: refractory hypoxaemia,
septic shock, multiorgan dysfunction syndrome, acute
myocardial infarction, acute heart failure, stroke, do-not-
resuscitate order, pulmonary thromboembolism, and
other; more than one cause of death could be considered.
Duration of mechanical ventilation and length of ICU
and hospital stay, in days, were also recorded.
Site investigators collected worst values for each variable
daily, from admission to the ICU to ICU discharge.
Definitions of comorbidities, physiological and mechanical
ventilation variables, complications, and causes of death
are provided in appendix 2 (pp 22–25).
 Articles

Outcomes Secondary outcomes were ICU mortality; independent

The primary outcome measure was all-cause in-hospital predictors of mortality; duration of invasive mechanical
mortality. All patients were followed until death or ventilation (in days); and patterns of change in
discharge, whichever occurred first. When the study was physiological respiratory and mechanical ventilation
designed and the protocol prepared (appendix 2, p 26) at variables on days 1 (ICU admission), 3, and 7 for the
the beginning of the pandemic, we selected all-cause entire cohort, and in survivor and non-survivor
ICU mortality as the primary outcome, similar to most subgroups. 28-day mortality was added as a modification
epidemiological studies. On Nov 15, 2020, we changed and was included as secondary outcome on Nov 15, 2020,
the outcome to in-hospital mortality, including patients for comparison with other reports.
who died in the ICU and after discharge from the ICU,
because it more compre­hensively reflects the full course Statistical analysis
of the disease. An additional, exploratory outcome was Before data analysis, two investigators (EE and FGR)
mortality and acuity over the duration of the study, screened the database for errors against standardised
from April to October, 2020, according to the month of ranges in each hospital. Investigators were contacted
hospital admission.12 (by EE and VSKE) with queries and to address
inconsistencies. Validated or corrected data were then
entered into the database. Missing data were not
All patients Survivors Non-survivors p value imputed. Given that this was an observational study
(n=1909) (n=808) (n=1101)
and there was no risk to patients, we sought to include
Age (years) 62 (52 to 70) 58 (49 to 68) 64 (55 to 72) <0∙0001 as many patients as possible, with no predefined
Sex sample size.
Male 1294 (67·8%) 537 (66·5%) 757 (68·8%) 0∙289 Variables are reported as absolute numbers and
Female 615 (32·2%) 271 (33·5%) 344 (31·2%) 0∙289 percentages, or medians and IQRs. Differences between
Weight (kg) 85 (75 to 100) 85 (75 to 100) 85 (75 to 98) 0∙930 survivors and non-survivors in recorded variables were
BMI (kg/m²) 29 (26 to 34) 29 (27 to 34) 29 (26 to 34) 0∙564 analysed with the χ² test or Fisher’s exact test, or the t test
Comorbidities or Wilcoxon rank-sum test, as appropriate. All tests were
Arterial hypertension 897/1909 (46·9%) 326/808 (40·3%) 571/1101 (51·9%) <0∙0001 two-sided, and a p value of <0·05 was considered to be
Obesity (BMI ≥30) 847/1909 (44·4%) 374/808 (46·3%) 473/1101 (43·0%) 0∙148 statistically significant.
Morbid obesity 290/1909 (15·2%) 123/808 (15·2%) 167/1101 (15·2%) 0∙974 Generalised estimating equations were used to account
(BMI >40 kg/m²) for correlations between respiratory variables in the
Diabetes 553/1909 (29·0%) 204/808 (25·2%) 349/1101 (31·7%) 0∙002 entire cohort over time, and between subgroups of
Respiratory disease 263/1909 (13·8%) 98/808 (12·1%) 165/1101 (15·0%) 0∙070 survivors and non-survivors. An unstructured correlation
Ischaemic heart disease 123/1909 (6·4%) 34/808 (4·2%) 89/1101 (8·1%) 0∙001 matrix was selected. p values for time-effect for the entire
Chronic kidney disease 118/1909 (6·2%) 33/808 (4·1%) 85/1101 (7·7%) 0∙001 cohort and for time-by-subgroup interaction were
Chronic heart failure 100/1909 (5·2%) 20/808 (2·4%) 80/1101 (7·3%) <0∙0001 calculated and a Bonferroni correction used to adjust for
Immunosuppression 92/1909 (4·8%) 28/808 (3·5%) 64/1101 (5·8%) 0∙018 multiple comparisons.
Oncohaematological disease 54/1909 (2·8%) 11/808 (1·4%) 43/1101 (3·9%) 0∙001 Mortality at 28 days and 90 days was plotted as time-to-
Chemotherapy in the 36/1909 (1·9%) 7/808 (0·1%) 29/1101 (2·6%) 0∙005 event curves using the Kaplan-Meier method. The Kaplan-
previous 6 months Meier analysis was cut at 90 days for simplicity, although
Chronic liver disease 34/1909 (1·8%) 7/808 (0·9%) 27/1101 (2·4%) 0∙010 some patients did not have an outcome (death or
Solid organ transplantation 17/1909 (0·9%) 5/808 (1·0%) 12/1101 (1·1%) 0∙279 discharge) at this point. Kaplan-Meier curves were also
Bone marrow 3/1909 (0·2%) 1/808 (0·1%) 20/1101 (1·8%) 1∙000 constructed to compare time-to-event differences in
transplantation
patients according to comorbidities (Charlson score <2
Pregnant or post-partum 4/1909 (0·2%) 3/808 (0·4%) 1/1101 (0·1%) 0∙317
and ≥2 points) and 10-year age category. Differences in
Presence of cardiovascular 944/1909 (49·4%) 336/808 (41·6%) 608/1101 (55·2%) <0∙0001
disease*
each case were analysed with the log-rank test.
Charlson comorbidity score 1 (1 to 2) 1 (0 to 2) 2 (1 to 3) <0∙0001
Cox regression analysis was used to determine
independent predictors of hospital mortality. Variables
No comorbidities 159/1909 (8·3%) 106/808 (13·1%) 53/1101 (4·8%) <0∙0001
differing between survivors and non-survivors with a
Habits and drug use
p value <0·20, according to the χ2 test or Fisher’s exact
ACE inhibitors or AII receptor 352/1909 (18·4%) 141/808 (17·5%) 211/1101 (19·2%) 0∙373
blockers test, or t test or Wilcoxon rank-sum test, were entered into
Current smoker 267/1909 (14·0%) 94/808 (11·6%) 173/1101 (15·7%) 0∙011 the multivariable regression model. Harrell’s C index was
Statins 137/1909 (7·1%) 58/808 (7·2%) 79/1101 (7·2%) 0∙998 calculated to test the predictive capacity of the model. The
β blockers 135/1909 (7·1%) 48/808 (5·3%) 87/1101 (7·9%) 0∙090
proportional hazard assumption was tested by visual
Self-reported alcohol-related 52/1909 (2·7%) 19/808 (2·4%) 33/1101 (3·0%) 0∙394
inspection of Schoenfeld residuals and by testing
problem predicted versus observed values of model variables.
(Table 1 continues on next page) For the post-hoc analysis of mortality over the months
of the study (April to October), a multiple χ² test was

4 www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0

 Articles

used. Age and acuity scores, such as APACHE II and

All patients Survivors Non-survivors p value
SOFA, were compared by means of one-way ANOVA. (n=1909) (n=808) (n=1101)
Missing data for each variable at each timepoint (day 1,
(Continued from previous page)
3, and 7) are shown in appendix 2 (pp 15–17). Data were
Duration of symptoms before 5 (3 to 7) 5 (3 to 7) 5 (3 to 7) 0∙266
analysed with Stata 14.0 (StataCorp LP, College Station,
admission to hospital (days)
TX, USA). The trial is registered with ClinicalTrials.gov,
Period between hospital and 1 (0 to 4) 1 (0 to 4) 1 (0 to 4) 0∙881
NCT04611269. ICU admission (days)
Respiratory management before ICU admission
Role of the funding source Prior use of non-invasive 73/1909 (3·8%) 24/808 (3·0%) 49/1101 (4·4%) 0·096
There was no funding source for this study. mechanical ventilation
Duration of non-invasive 1 (1 to 2) 1 (1 to 2) 1 (1 to 2) 0·37
Results mechanical ventilation
(days)
Between March 20, 2020, and Oct 31, 2020, we enrolled
Prior use of high-flow nasal 144/1909 (7·5%) 58/808 (7·2%) 86/1101 (7·8%) 0·61
1909 invasively ventilated patients with RT-PCR-con­ cannula
firmed COVID-19, admitted to 63 ICUs (figure 1). Patients
Duration of high-flow nasal 1 (0 to 2) 1 (0 to 2) 1 (0 to 2) 0·14
had a median age of 62 years (IQR 52–70), were cannula use (days)
predominantly male (1294 [67·8%]), and 1750 (91·7%) Requirement for invasive 129/1909 (6·8%) 40/808 (5·0%) 89/1101 (8·1%) 0∙01
had comorbidities, of which arterial hypertension mechanical ventilation
(897 [46·9%]), obesity (847 [44·4%]), and diabetes before ICU admission
(553 [29·0%]) were the most frequent (table 1). Patients Duration of invasive 1 (1 to 2) 1 (1 to 2) 1 (1 to 2) 0∙460
mechanical ventilation
stayed in hospital after admission for a median of 1 day before ICU admission (days)
(IQR 0–4) before being admitted to the ICU. 144 (7·5%) Endotracheal intubation 416/1872 (22·2%) 150/789 (19·0%) 266/1083 (24·6%) 0·004
patients were on HFNC for a median of 1 day (0–2) and outside the ICU
73 (3·8%) patients were on NIV for a median of 1 day (1–2). Variables of disease severity†
416 (22·2%) of 1872 patients underwent endotracheal APACHE II 15 (10 to 20) 13 (9 to 18) 16 (12 to 22) <0∙0001
intubation outside the ICU, 283 (15·1%) in the emergency SOFA24-h 5 (3 to 8) 4 (3 to 7) 6 (4 to 8) <0∙0001
department and 133 (7·1%) in the general ward (table 1). Pre-intubation respiratory 32 (28 to 36) 32 (28 to 36) 32 (28 to 36) 0·530
Symptoms lasted for a median of 5 days (3–7) before rate
hospital admission, the most common of which were Oxygen saturation by pulse 89 (86 to 94) 91 (88 to 94) 89 (85 to 93) <0∙0001
dyspnoea (1443 [75·6%]), fever (1424 [74·6%]), and cough oximetry at admission
(1188 [62·2%]; appendix 2, p 8). Extension of lung infiltrates 1324/1650 (80·2%) 553/701 (78·9%) 771/949 (81·2%) 0∙235
over 3–4 quadrants on chest
In patients admitted to the ICU, 1456 (79·3%) with x-ray or CT scan
spontaneous breathing had tachypnoea, median oxygen
Requirement for 939/1909 (49·2%) 345/808 (42·7%) 594/1101 (53·9%) <0∙0001
saturation was 89% (IQR 86–94), 1324 (80·2%) had lung vasopressors
infiltrates involving 3–4 quadrants on chest x-ray, and Fluid balance on day 1 (mL) 716 650 787 0∙310
these 1456 patients underwent endotracheal intubation ( to 100 to 1700) ( to 53 to 1600) ( to 154 to 1779)
within 0 days (0–1) from admission to the ICU. Data are n/N (%) or median (IQR). Percentages were calculated according to the data recorded for each variable. Missing
Requirement for vasopressors was common (939 [49·2%]). data corresponding to each variable are shown in appendix 2 (pp 15–16). ACE=angiotensin-converting enzyme.
The most frequent laboratory alterations were mild AII=angiotensin II. APACHE II=Acute Physiology and Chronic Health Evaluation. BMI=body-mass index. ICU=intensive care
unit. SOFA=Sequential Organ Failure Assessment. *Includes any cardiovascular disease: arterial hypertension, ischaemic
leucocytosis with lymphopenia, and increased lactate heart disease, chronic heart failure. †Calculated within the first 24 h of ICU admission.
dehydrogenase and markers of inflammation, such as
D-dimer, ferritin, and arterial lactate con­ centrations Table 1: Epidemiological variables and risk factors in invasively ventilated patients with COVID-19
(table 2).
Median duration of invasive mechanical ventilation
was 13 days (IQR 7–22). Physiological respiratory and trends were noted for blood pH, bicarbonate, and base
mechanical ventilation variables in the entire cohort on excess. Hypercapnia was present in all patients from day
days 1, 3, and 7 are shown in appendix 2 (pp 9–10). 1 and remained stable. 1779 (97·4%) of 1827 patients were
Median tidal volume administered was 6∙1 mL/kg receiving deep sedation (RASS of –4 or –5 points) on
predicted bodyweight (IQR 6∙0–7∙0) on day 1, and the day 1, but this proportion decreased over time.
value increased significantly up to day 7; applied PEEP 1101 (57∙7%) of 1909 patients died in hospital (primary
levels were intermediate at 10 cm H2O (8–12) on day 1, outcome). 28-day mortality was 50∙6% (966 of
with a slight but significant decrease to day 7. PaO2/FiO2 1909 patients) and ICU mortality was 57∙0% (1088 of
was 160 (111–218) on day 1, increased to day 3, and 1909 patients). Kaplan-Meier survival estimates for the
stabilised by day 7. Median respiratory system compliance entire group, and according to Charlson score (<2 and
(36 mL/cm H2O [29–44] on day 1), plateau pressure ≥2 points) and age category, are shown in
(23 cm H2O [20–26]), and FiO2 (0∙60 [0∙45–0∙80]) appendix 2 (pp 19–21). Mortality increased with age
improved slightly but significantly over time; similar category (figure 2A) and over the study period, from

www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0 5

 Articles

survivors than in non-survivors. Tidal volume increased

All patients Survivors Non-survivors p value
(n=1909) (n=808) (n=1101) significantly over time in both groups (from day 1 to
day  7), although the increase was less in non-survivors
Haemoglobin (g/L) 13 (11–14) 13 (12–14) 13 (11–14) 0·001
than survivors. PEEP levels were similar in both
White blood cell count 11·0 (7·6–15·0) 10·4 (7·2–14·2) 11·4 (7·9–15·5) <0·0001
(× 10⁹ per L) subgroups. At all timepoints, variables for respiratory
Lymphocyte count 0·8 (0·5–1·1) 0·8 (0·5–1·1) 0·7 (0·5–1·1) 0·110
mechanics, such as respiratory system compliance,
(× 10⁹ per L) plateau pressure, driving pressure, and FiO2, showed
Platelet count 224·0 (168·0–299·0) 224·0 (171·0–302·0) 224·0 (166·0–294·0) 0·190 significant differences between survivors and non-
(× 10⁹ per L) survivors. The proportion of patients requiring deep
Aspartate 42 (29–65) 41 (29–65) 43 (28–66) 0·505 sedation decreased in survivors after day 1; it also
aminotransferase (U/L) decreased in non-survivors but to a lesser extent.
Alanine 39 (26–64) 40 (27–69) 39 (24–63) 0·009 Complications in all patients were frequent. ARDS
aminotransferase (U/L)
developed in 1672 (87·6%) patients. Prone positioning
Total bilirubin (μmol/L) 10·3 (6·8–15·4) 9·4 (6·8–15·4) 10·3 (6·8–15·7) 0·006
was used in 1176 (61·6 %) patients, most frequently in
Lactate dehydrogenase 512 (355–750) 453 (326–653) 558 (383–829) <0·0001
(U/L)
non-survivors, who received more sessions (table 3).
Blood urea nitrogen 3·42 (2·41–5·28) 2·95 (2·17–4·20) 4·04 (2·72–5·98) <0·0001
Extracorporeal membrane oxygenation was seldom used
(mmol/L) (n=1, survivor). Other complications, such as septic shock,
Creatinine (μmol/L) 79·6 (61·9–114·9) 70·7 (61·9–97·2) 88·4 (61·9–132·6) <0·0001 acute kidney injury, and requirement for renal
D-dimer (mg/L) 1·13 (0·56–3·08) 0·90 (0·50–2·21) 1·44 (0·70–3·83) <0·0001 replacement therapy were common, while thrombotic
Ferritin (ng/mL) 1063 (545–1775) 1031 (525–1885) 1108 (580–1700) 0·493 episodes developed in only 170 (8·9%) patients; all were
Arterial lactate 1·8 (1·4–2·4) 1·7 (1·3–2·3) 1·9 (1·5–2·6) <0·0001 significantly more frequent in non-survivors than
(mmol/L) survivors (table 3). Corticosteroid administration did not
Data are expressed as median (IQR). Numbers and percentages of missing data, and their distribution across survivors differ between subgroups, but its use increased
and non-survivors, are shown in appendix 2 (pp 15–16). significantly from 576 (68·2%) of 844 patients in the
April–July period to 1006 (96·1%) of 1047 patients in the
Table 2: Laboratory findings
August–October period. Infectious complications were
similar in both subgroups. One-quarter of patients
April to October (except for a small, non-significant underwent tracheostomy, which was more frequent in
decrease observed in June; figure 2B). Specifically, in an survivors. Duration of invasive mechanical ventilation, as
exploratory, post-hoc analysis, mortality was significantly well as ICU and hospital stays, were prolonged, although
higher for patients admitted during the period from shorter in non-survivors (table 3). Refractory hypoxaemia
August to October, the second half of the study, compared was the single most common cause of mortality in the
with April to July (figure 2B). However, patient acuity was entire cohort, followed by septic shock and multiorgan
higher during the first month of the study than in the dysfunction syndrome, and patients often had more than
subsequent months of May to October (appendix 2, p 11). one cause of death (appendix 2, p 13).
Older age, higher Charlson score, cardiovascular Cox regression identified age (hazard ratio [HR] 1∙02
disease, chronic kidney disease, immunosuppression, [95% CI 1∙01–1∙03]), Charlson score (1∙16 [1∙11–1∙23]),
smoking, arterial hypertension, diabetes, ischaemic endotracheal intubation outside the ICU (1∙37
heart disease, chronic heart failure, oncohaematological [1∙10–1∙71]), vasopressor use on day 1 (1∙29 [1∙07–1∙55]),
disease, and having received chemotherapy were D-dimer concentration (1∙02 [1∙01–1∙03]), PaO2/FiO2 on
significantly more common in non-survivors (n=1101) day 1 (0∙998 [0∙997–0∙999]), arterial pH on day 1 (1∙01
than in survivors (n=808; table 1). Non-survivors were [1∙00–1∙01]), driving pressure on day 1 (1∙05 [1∙03–1∙08]),
sicker on admission than survivors, with significantly acute kidney injury (1∙66 [1∙36–2∙03]), and month of
higher APACHE II and SOFA scores, and higher use of admission (1∙10 [1∙03–1∙18]) as independent predictors
vasopressors, and they were more likely to receive of mortality (appendix 2, p 14).
invasive mechanical ventilation before ICU admission
(table 1). Similarly, most laboratory values differed Discussion
significantly between non-survivors and survivors at SATICOVID was a large, prospective cohort study of
admission―even non-survivors with laboratory variables 1909 patients with COVID-19 requiring invasive mechanical
within the healthy range were still significantly different ventilation and ICU admission in Argentina, an upper
to survivors (table 2).
 Differences in physiological middle-income country. Patients were predominantly older
respiratory and mechanical ventilation variables between and male, had symptoms for a median of 5 days before
survivors and non-survivors on days 1, 3, and 7 are shown admission to hospital, and stayed a median of 1 day in
in figure 3 and in appendix 2 (pp 9, 18). Throughout the hospital before being admitted to the ICU. 11·4% (217 of
study period (on days 1, 3, and 7), PaO2/FiO2, blood pH, 1909 patients) of patients had not improved on HFNC or
and base excess were significantly higher, and PaCO2 and NIV, and 22·2% underwent endotracheal intubation
lactate concent­ rations were significantly lower, in outside the ICU.

6 www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0

 Articles

A B
100

80 †
† §
† ‡
Mortality (%)

60
*
40

20

0
<40 40–49 50–59 60–69 70–79 ≥80 April May June July August September October
(n=156) (n=224) (n=422) (n=587) (n=395) (n=121) (n=68) (n=88) (n=248) (n=440) (n=549) (n=374) (n=132)
Age (years) Month of admission

Figure 2: All-cause in-hospital mortality according to age category (A) and month of hospital admission (B)
Number of patients in each category included on x axis. *Indicates p=0·007 vs age <40 years. †p<0·0001 vs age <40 years. ‡p=0·018 vs April. §p=0·005 vs April.

Survivors Non-survivors
A B C * * *
D
12 * 50 1·00 25
* *
10 40 0·80 20
Respiratory rate

PEEP (cm H2O)

(breaths/min)

8
Tidal volume

30 0·60 15
(mL/kg)

6
FiO2

20 0·40 10
4
2 10 0·20 5
0 0 0·00 0

E F G * * * H * * *
80 150 30 600
* * *
Plateau pressure

60
Static compliance

Driving pressure

* *
(mL/cm H₂O)

* 100 20 400
(cm H2O)

PaO2/FiO₂
(cm H₂O)

40

20 50 10 200

0
0 0 0
Day 1 Day 3 Day 7 Day 1 Day 3 Day 7 Day 1 Day 3 Day 7 Day 1 Day 3 Day 7

Figure 3: Lung mechanics, mechanical ventilation, and PaO2/FiO2 in survivors and non-survivors, at days 1, 3, and 7
A time-by-group interaction is present for all variables (p<0·001) except for PEEP levels. FiO2=fractional concentration of oxygen in inspired air. PaO2/FiO2=ratio of partial pressure of arterial oxygen to
fractional inspired oxygen. PEEP=positive end-expiratory pressure. *Differences between survivors and non-survivors, when present (p<0·01), are given for each timepoint, corrected for multiple
comparisons.

Overall in-hospital mortality was high, at 57∙7%. High LMICs showed higher mortality for sepsis and ARDS
mortality for invasively ventilated patients has been seen than did high-income countries.19,20 Complex economic
in studies from China (49%), Lombardy, Italy (53%), and and organisational factors in LMICs explain worse
Germany (55%).7,9,13,14 Patients in the German study7 were outcomes for ICU patients. Deep inequities, defined as
about 6 years older than those in our study, while patients systematic, unjust, and preventable differences in
from Italy were of a similar age to ours (63 years);9 in the determinants of health, such as socioeconomic status,
Chinese study,13 the median age of critical cases was not demographics, and geography, might generate
shown. However, in patients 70 years and older, mortality differences in access to health services in different
was similar in Germany and in our study; for example, in population subgroups, which affect health-related
patients aged 70–79 years, mortality was 63% in the outcomes.3 Furthermore, in LMICs, health systems are
German study7 and 68% in ours, and in those aged usually fragmented in public, private, and social security
80 years and older, mortality was 72% and 75%, sectors, which maintain the differences according to
respectively. Conversely, other cohorts have shown lower socioeconomic status and affect the provision of health
mortality rates in patients receiving invasive mechanical care, particularly critical care.21
ventilation, such as 28% in New York, 31% in France, Identifying independent determinants of prognosis in
32% in Spain, 35% in the Netherlands, and 43% in the critically ill, mechanically ventilated patients with
UK.12,15–18 In two large, retrospective population studies in COVID-19 is key to optimising use of ICU resources. As
Mexico and Brazil, in-hospital mortality of patients with in other cohorts of patients with COVID-19, increasing
COVID-19 on mechanical ventilation was high age was an independent predictor of mortality. Risk
(76% and 80%, respectively).5,6 These differences mirror factors for mortality were similar to those identified in
findings in the ICON and LUNG SAFE studies, in which other studies, in general, but the presence of at least one

www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0 7

 Articles

All patients Survivors Non-survivors p value

(August–October), were also independently associated
(n=1909) (n=808) (n=1101) with mortality.
ARDS development 1672/1909 (87·6%) 658/808 (81·4%) 1014/1101 (92·1%) <0·0001
Intubation of patients outside the ICU might reflect
Prone positioning 1176/1909 (61·6%) 430/808 (53·2%) 746/1101 (67·8%) <0·0001
severity of hypoxaemia on admission to hospital or rapid
Number of sessions 2 (2–4) 2 (1–4) 3 (2–4) 0·041
deterioration on the general ward. It might also indicate
insufficient number of ICU beds; however, patients were
Duration of sessions (h) 24 (21–36) 24 (20–35) 24 (22–36) 0·051
on invasive mechanical ventilation outside the ICU for
Septic shock 1513/1909 (79·3%) 539/808 (63·7%) 974/1101 (88·5%) <0·0001
only a short duration before ICU admission. Conversely,
Acute kidney injury 997/1909 (52·2%) 272/808 (32·7%) 725/1101 (65·8%) <0·0001
intensivists’ experience with timely management of
Renal replacement therapy 373/1909 (19·5%) 91/808 (11·3%) 282/1101 (25·6%) <0·0001
severe respiratory failure might have contributed to
Ventilator-associated 617/1909 (32·3%) 267/808 (33·0%) 350/1101 (31·8%) 0·54
pneumonia decreased mortality in patients intubated in the ICU
Bacteraemia (all 446/1909 (23·4%) 192/808 (23·8%) 254/1101 (23·1%) 0·72
compared with those intubated before ICU admission.
microorganisms) The reduction in PaO2/FiO2 ratio was similar to that
Bacteraemia (Gram-negative 191/1909 (10·0%) 81/808 (10·0%) 110/1101 (10·0%) 0·98 seen in studies of patients with COVID-19 from Italy,
bacilli) France, Spain, and the Netherlands.9,12,16,17 As in these
Maximum fever (°C) 38·5 (38·0–39·0) 38·5 (38·0–39·0) 38·6 (38·0–39·0) 0·20 reports, compliance with lung-protective ventilation was
Maximum fever ≥39°C 419/1832 (22·9%) 162/599 (32·1%) 257/783 (32·8%) 0.02 high. Tidal volumes used were between 6∙1 and 6∙5 mL/kg
Thromboembolic 170/1909 (8·9%) 57/808 (7·1%) 113/1101 (10·3%) 0·02 predicted bodyweight, except in survivors at day 7; plateau
complications pressures were less than 30 cm H2O, and driving pressures
Deep vein thrombosis 57/1909 (3·0%) 28/808 (3·4%) 29/1101 (2·6%) 0·29 were less than 15 cm H2O at day 1, 3, and 7.9,12,16,17 FiO2
Pulmonary embolism 62/1909 (3·2%) 24/808 (3·0%) 38/1101 (3·5%) 0·25 remained between 0∙45 and 0∙60; however, PEEP values
Ischaemic stroke 17/1909 (0·9%) 3/808 (1·0%) 14/1101 (1·3%) 0·03 (approximately 10 cm H2O) were slightly lower than those
Ischaemia of the 14/1909 (0·7%) 2/808 (0·2%) 12/1101 (1·1%) 0·03 previously reported for patients with COVID-19.9,12,16,17 A
extremities
similar use of protective mechanical ventilation has been
Dexamethasone use 1612/1909 (84·4%) 662/800 (82·8%) 950/1101 (86·3%) 0·09
reported from Asian middle-income countries.24 In
Convalescent plasma use 605/1909 (31·7%) 305/808 (37·8%) 300/1101 (27·3%) <0·0001
accordance with existing information for ARDS not
Tracheostomy 464/1909 (24·3%) 293/808 (36·2%) 171/1101 (15·5%) <0·0001 related to COVID-19,25 driving pressure was strongly
Length of invasive mechanical 13 (7–22) 14 (8–28) 11 (6–19) <0·0001 associated with mortality, which is a novel finding.
ventilation (days)
Oxygenation and mechanical ventilation variables
Length of ICU stay (days) 16 (10–27) 23 (14–37) 13 (8–21) <0·0001
differed consistently between survivors and non-survivors
Length of hospital stay (days) 22 (13–35) 34 (23–51) 16 (10–24) <0·0001
over time. Of note, tidal volume and PEEP were similar
Data are n/N (%) or median (IQR). Numbers and percentages of missing data, and their distribution across survivors in both subgroups at all time points (with the exception
and non-survivors, are shown in appendix 2 (pp 15–16). ARDS=acute respiratory distress syndrome. ICU=intensive care
unit.
of tidal volume at day 7). Since refractory hypoxaemia
was the most common cause of death, there is probably
Table 3: Disease evolution, therapeutic modalities, and complications during ICU stay some room for improvement in the FiO2 and PEEP
settings used for patients. Prone positioning, which is
associated with better outcomes in ARDS,26 was used in
comorbidity in 92% of our patients is, to our knowledge, most patients, as in other COVID-19 studies.9,15,27 This
the highest recorded.5,6,9,12,18 Obesity was highly prevalent practice was more frequent in non-survivors, probably
in our cohort (44·4%) but, surprisingly, it was not reflecting its application in the case of the most severely
associated with increased mortality, as reported in other affected patients.
studies.22 However, another study found that mortality With respect to haemodynamic alterations, our study
did not differ between BMI categories.23 Other conditions, highlights the relevance of cardiovascular dysfunction.
such as arterial hypertension, diabetes, chronic kidney The requirement of vasopressors by patients on day 1
failure, cardiovascular disease, and immunosuppression was frequent and was independently associated with
were more frequent in non-survivors. No single hospital mortality, even with lactate levels of 2∙0 mmol/L
comorbidity was independently associated with mortality or less. Accordingly, reports have identified cardiovascular
in the Cox regression analysis (data not shown), but SOFA score and the lowest systolic blood pressure
Charlson score, a validated comorbidity index, was recorded for a patient as predictors of mortality.12,28
independently associated with mortality. Renal dysfunction in COVID-19 is common and can
Although the effect of older age and pre-existing occur via various mechanisms, but its development in
conditions on mortality is clear, it is worth noting that patients on invasive mechanical ventilation implies a
other factors, such as profound physiological derange­ worse prognosis.29,30 In our study, blood urea nitrogen and
ments on day 1 (eg, decreased oxygenation, increased serum creatinine concentrations were already significantly
driving pressure, requirement for vasopressors, acidosis, different between survivors and non-survivors on
activation of coagulation), intubation outside the ICU, admission at day 1, and acute kidney injury during the ICU
and admission during the second half of the study stay was a strong, independent predictor of mortality. The

8 www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0


activation of both thrombotic and fibrinolytic pathways,
reflected by increased D-dimer values in patients admitted
to hospital with COVID-19, was reported early in the
pandemic and has been independently associated with
mortality.31 Our study confirms these findings, which are
also in line with the findings of a meta-analysis.32
The month of hospital admission was independently
associated with mortality, but whereas mortality was
reported to improve over time in France and the UK, we
found that mortality was higher among patients admitted
in the later months compared with April, in Argentina.12,18
This increase in mortality cannot be ascribed to differences
in age or in the severity of disease, because patients were
more severely ill on admission to hospital during the first
month of the pandemic. Nor can it be attributed to low
adherence to the only therapeutic measure proven to be
effective (dexamethasone); on the contrary, administration
of corticosteroids increased after RECOVERY trial results
were published in July, 2020.11 We believe that the increase
in mortality over time might reflect the profound stress
placed on the health system by the pandemic, counteracting
the benefits of learning related to the management of
COVID-19 over the study period. In Argentina, ICU beds,
ventilators, and personal protective equipment were widely
available in periods of increased ICU demand due to
timely acquisition and distribution by the government and
by private and non-profit organisations. However, ICU
personnel became scarce. The number of intensivists was
already low before the pandemic, and many contracted
COVID-19 or even died as the peak of cases approached.33
Although the health system was not overwhelmed in terms
of insufficient equipment, denial of care, or a lack of beds,
lower quality of care might have occurred because of the
high and sustained burden on health-care personnel.
Evidence for an effect of increased ICU strain on health-
care workers on mortality has been reported in a
retrospective analysis of 8516 patients with COVID-19
admitted to 88 US Veterans Affairs hospitals.34 Patients
who were treated during periods of peak ICU demand had
nearly twice the risk of mortality compared with patients
treated during periods of low demand. Moreover, the
duration of mechanical ventilation in the ICU and ICU
and hospital stays were prolonged over the course of the
study, as described in other cohorts, which certainly
contributed to the burden on the health-care system.6,12,17
This study has several strengths. It was conducted
prospectively, and it is one of the largest cohorts of
patients with COVID-19 requiring invasive mechanical
ventilation. It provides a comprehensive evaluation of
risk factors, markers of disease severity, patterns of
change in respiratory variables, use of lung-protective
strategies, complications, causes of death, and prognostic
factors. It is, to our knowledge, the first exhaustive Latin
American study in a setting of scarce information about
the most severely affected patients with COVID-19 in
LMICs. RT-PCR testing is standardised in Argentina,
which makes diagnosis homogeneous. All patients in the
www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0 9
Articles
study completed the course of disease to death or hospital
discharge.
Nevertheless, this study has some limitations. First,
since participation in the study was voluntary, ICUs with
higher or lower mortality might be under-represented,
and the final figure we report for in-hospital mortality
might be different to that of unselected, nationwide
cohorts in Latin America. Second, admission policies and
patient management might have differed between the
centres in our study. Third, non-ventilated patients with
COVID-19 in the ICU were not included, so the full
spectrum of disease was not characterised. Fourth,
notwithstanding the prospective nature of the study, some
variables have missing data due to the high burden of
work during the pandemic and the scarce time personnel
had available to collect data (as reported in other studies).
Nevertheless, in the case of most variables, data were
missing for less than 5% of cases. Exceptions were
D-dimer, lactate, and ferritin concentrations due to lack of
laboratory capacity for their measurement in some
centres. Fifth, to minimise the workload for health-care
workers in the ICU, data registration beyond the date of
admission to the ICU (ie, day 1), or beyond day 7 for
ventilation management, was not performed. Therefore,
we cannot exclude an effect of these unrecorded variables
on mortality. Sixth, five centres recruited fewer than
five patients and some patients with COVID-19 might
have been missed due to the lack of personnel. Finally,
data collected in Argentina might not be representative of
other LMICs and other regions.
To conclude, in SATICOVID, in-hospital mortality was
high in patients with COVID-19 requiring invasive
mechanical ventilation. Pre-existing conditions, such as
age and Charlson index, together with physiological
impairments (alterations in oxygenation, presence of
hypotension, acidosis, acute kidney injury, and activation
of coagu­ lation) and mechanical ventilation variables,
were independent predictors of in-hospital mortality.
Thus, signs of early organ dysfunction (ie, alterations in
oxygenation, presence of hypotension, acidosis, acute
kidney injury, and activation of coagulation) appear to be
a prognostic factor in severe COVID-19. We also found a
paradoxical increase in mortality throughout the first
wave of the pandemic, possibly reflecting increasing
strain on the health-care system. Long duration of
mechanical ventilation and prolonged ICU stay
contributed to the pressure on ICU capacity. We believe
that the information provided here will help to improve
health-care management in the second wave of the
pandemic and beyond.
Contributors
EE, GP, RR, FGR, and VSKE conceived and designed the study. EE, CIL,
and AD analysed the data. EE and AD drafted the manuscript. FGR, GP,
and VSKE were in charge of the project administration. AD designed the
figures. EE, AD, and CIL verified the data. EE, GP, RR, FGR, VSKE, CIL,
AD, MA, IR, DP, MB, VM, CG, ST, CO, PNRB, MFV, EC, MGS, NT,
and VA contributed to the acquisition and interpretation of data.
All authors had full access to all the data and had responsibility for the
 Articles
decision to submit for publication. All authors have seen and approved
the final version of the manuscript.
Declaration of interests
The authors declare no competing interests.
Data sharing
De-identified individual participant data that underlie the results
reported in this Article (text, tables, figures, and appendices),
data dictionaries, and study protocol will be available from 9 months to
36 months after Article publication to researchers who provide a
methodologically sound proposal, for any purpose of analysis. Proposals
should be directed to estenssoro.elisa@gmail.com; to gain access,
data requestors will need to sign a data access agreement.
References
1 WHO. WHO coronavirus disease (COVID-19) dashboard.
https://covid19.who.int/ (accessed Feb 9, 2021).
2 Kirby T. South America prepares for the impact of COVID-19.
Lancet Respir Med 2020; 8: 551–52.
3 Estenssoro E, Loudet CI, Edul VSK, et al. Health inequities in the
diagnosis and outcome of sepsis in Argentina: a prospective cohort
study. Crit Care 2019; 23: 250.
4 The Lancet. COVID-19 in Latin America: a humanitarian crisis.
Lancet 2020; 396: 1463.
5 Ñamendys-Silva SA, Gutiérrez-Villaseñor A, Romero-González JP.
Hospital mortality in mechanically ventilated COVID-19 patients in
Mexico. Intensive Care Med 2020; 46: 2086–88.
6 Ranzani OT, Bastos LSL, Gelli JGM, et al. Characterisation of the first
250 000 hospital admissions for COVID-19 in Brazil: a retrospective
analysis of nationwide data. Lancet Respir Med 2021; 407–18.
7 Karagiannidis C, Mostert C, Hentschker C, et al. Case
characteristics, resource use, and outcomes of 10 021 patients with
COVID-19 admitted to 920 German hospitals: an observational
study. Lancet Respir Med 2020; 8: 853–62.
8 Estenssoro E, Ríos FG, Apezteguía C, et al. Pandemic 2009
influenza A in Argentina: a study of 337 patients on mechanical
ventilation. Am J Respir Crit Care Med 2010; 182: 41–48.
9 Grasselli G, Zangrillo A, Zanella A, et al. Baseline characteristics
and outcomes of 1591 patients infected with SARS-CoV-2 admitted
to ICUs of the Lombardy region, Italy. JAMA 2020; 323: 1574–81.
10 Ranieri VM, Rubenfeld GD, Thompson BT, et al. Acute respiratory
distress syndrome: the Berlin Definition. JAMA 2012; 307: 2526–33.
11 Horby P, Lim WS, Emberson JR, et al. Dexamethasone in
hospitalized patients with Covid-19. N Engl J Med 2021; 384: 693–704.
12 COVID-ICU Group on behalf of the REVA Network and the
COVID-ICU Investigators. Clinical characteristics and
day-90 outcomes of 4244 critically ill adults with COVID-19:
a prospective cohort study. Intensive Care Med 2021; 47: 60–73.
13 Wu Z, McGoogan JM. Characteristics of and important lessons
from the coronavirus disease 2019 (COVID-19) outbreak in China:
summary of a report of 72 314 cases from the Chinese Center for
Disease Control and Prevention. JAMA 2020; 323: 1239–42.
14 Grasselli G, Greco M, Zanella A, et al. Risk factors associated with
mortality among patients with COVID-19 in intensive care units in
Lombardy, Italy. JAMA Intern Med 2020; 180: 1345–55.
15 Richardson S, Hirsch JS, Narasimhan M, et al. Presenting
characteristics, comorbidities, and outcomes among 5700 patients
hospitalized with COVID-19 in the New York City area. JAMA 2020;
323: 2052–59.
16 Ferrando C, Suarez-Sipmann F, Mellado-Artigas R, et al. Clinical
features, ventilatory management, and outcome of ARDS caused by
COVID-19 are similar to other causes of ARDS. Intensive Care Med
2020; 46: 2200–11.
17 Botta M, Tsonas AM, Pillay J, et al. Ventilation management and
clinical outcomes in invasively ventilated patients with COVID-19
(PRoVENT-COVID): a national, multicentre, observational cohort
study. Lancet Respir Med 2021; 9: 139–48.
10 www.thelancet.com/respiratory Published online July 2, 2021 https://doi.org/10.1016/S2213-2600(21)00229-0
18 Doidge JC, Gould DW, Ferrando-Vivas P, et al. Trends in intensive
care for patients with COVID-19 in England, Wales and Northern
Ireland. Am J Respir Crit Care Med 2021; 203: 565–74.
19 Vincent JL, Marshall JC, Namendys-Silva SA, et al. Assessment of
the worldwide burden of critical illness: the intensive care over
nations (ICON) audit. Lancet Respir Med 2014; 2: 380–86.
20 Laffey JG, Madotto F, Bellani G, et al. Geo-economic variations in
epidemiology, patterns of care, and outcomes in patients with acute
respiratory distress syndrome: insights from the LUNG SAFE
prospective cohort study. Lancet Respir Med 2017; 5: 627–38.
21 Estenssoro E, Alegria L, Murias G, et al. Organizational issues,
structure and processes of care in 257 ICUs in Latin America:
a study of the Latin America Intensive Care Network. Crit Care Med
2017; 45: 1325–36
22 Popkin BM, Du S, Green WD, et al. Individuals with obesity and
COVID-19: a global perspective on the epidemiology and biological
relationships. Obes Rev 2020; 21: e13128.
23 Schavemaker R, Schultz MJ, Lagrand WK, van Slobbe-Bijlsma ER,
Serpa Neto A, Paulus F. Associations of body mass index with
ventilation management and clinical outcomes in invasively
ventilated patients with ARDS related to COVID-19-insights from
the PRoVENT-COVID Study. J Clin Med 2021; 10: 1176.
24 Pisani L, Algera AG, Serpa Neto A, et al. Epidemiological
characteristics, ventilator management, and clinical outcome in
patients receiving invasive ventilation in intensive care units from
10 Asian middle-income countries (PRoVENT-iMiC):
an international, multicenter, prospective study. Am J Trop Med Hyg
2021; 104: 1022–33.
25 Bellani G, Laffey JG, Pham T, et al. Epidemiology, patterns of care,
and mortality for patients with acute respiratory distress syndrome
in intensive care units in 50 countries. JAMA 2016; 315: 788–800.
26 Guérin C, Reignier J, Richard JC, et al. Prone positioning in severe
acute respiratory distress syndrome. N Engl J Med 2013;
368: 2159–68.
27 Ferrando-Vivas P, Doidge J, Thomas K, et al. Prognostic factors for
30-day mortality in critically ill patients with coronavirus disease
2019: an observational cohort study. Crit Care Med 2021; 49: 102–11.
28 Jimenez-Solem E, Petersen TS, Hansen C, et al. Developing and
validating COVID-19 adverse outcome risk prediction models from
a bi-national European cohort of 5594 patients. Sci Rep 2021;
11: 3246.
29 Cheng Y, Luo R, Wang K, et al. Kidney disease is associated with
in-hospital death of patients with COVID-19. Kidney Int 2020;
97: 829–38.
30 Chaibi K, Dao M, Pham T, et al. Severe acute kidney injury in
patients with COVID-19 and acute respiratory distress syndrome.
Am J Respir Crit Care Med 2020; 202: 1299–301.
31 Zhou F, Yu T, Du R, et al. Clinical course and risk factors for
mortality of adult inpatients with COVID-19 in Wuhan, China:
a retrospective cohort study. Lancet 2020; 395: 1054–62.
32 Gungor B, Atici A, Baycan OF, et al. Elevated D-dimer levels on
admission are associated with severity and increased risk of
mortality in COVID-19: a systematic review and meta-analysis.
Am J Emerg Med 2021; 39: 173–79.
33 Ministry of Health of Argentina. Analysis of health-care personnel
affected by COVID-19. https://www.argentina.gob.ar/noticias/
analisis-de-la-situacion-del-personal-de-salud-afectado-por-covid-19
(accessed June 21, 2021).
34 Bravata DM, Perkins AJ, Myers LJ, et al. Association of intensive
care unit patient load and demand with mortality rates in US
Department of Veterans Affairs Hospitals during the COVID-19
pandemic. JAMA Netw Open 2021; 4: e2034266.